Participant Fluctuation Amplitude Data

fluctuations.glasser.pnc <- read.csv("/cbica/projects/spatiotemp_dev_plasticity/FluctuationAmplitude/PNC/glasserfluctuations_demographics_finalsample.csv") #fMRI + demographics, generated with fitGAMs_fluctuationamplitude_age.R

Glasser Labels

glasser.parcel.labels <- read.csv("/cbica/projects/spatiotemp_dev_plasticity/Maps/parcellations/surface/glasser360_regionlist.csv", header = T) #glasser region list

Sensorimotor-Association Axis

S.A.axis.cifti <- read_cifti("/cbica/projects/spatiotemp_dev_plasticity/Maps/S-A_ArchetypalAxis/FSLRVertex/SensorimotorAssociation_Axis_parcellated/SensorimotorAssociation.Axis.Glasser360.pscalar.nii") #S-A_ArchetypalAxis repo, vertex-wise axis average ranks
S.A.axis <- as.data.frame(cbind(rank(S.A.axis.cifti$data), names(S.A.axis.cifti$Parcel)))
colnames(S.A.axis) <- c("SA.axis","orig_parcelname")
S.A.axis <- merge(S.A.axis, glasser.parcel.labels, by="orig_parcelname", sort = F)
S.A.axis$SA.axis <- as.numeric(S.A.axis$SA.axis)
rm(S.A.axis.cifti)

Intracortical Myelination Developmental Data

myelin.partialR2.cifti <- read_cifti("/cbica/projects/spatiotemp_dev_plasticity/Myelin/hcpd_n628_myelin_sAge_partial_bayes_r2.pscalar.nii") #T1/T2 ratio - age effect size
myelin.maxdev.cifti <- read_cifti("/cbica/projects/spatiotemp_dev_plasticity/Myelin/hcpd_n628_median_posterior_age_of_max_slope_myelination.pscalar.nii") #age of maximal T1/T2 ratio increase
myelin <- as.data.frame(cbind(myelin.partialR2.cifti$data, myelin.maxdev.cifti$data))
colnames(myelin) <- c("myelin.partialR2","myelin.maxdev")
myelin$label <- glasser.parcel.labels$label
rm(myelin.partialR2.cifti)
rm(myelin.maxdev.cifti)

SNR Mask

SNR.mask <- read.csv("/cbica/projects/spatiotemp_dev_plasticity/Maps/parcellations/surface/SNRmask_glasser360.csv")

Spin Test Parcel Rotation Matrix

source("/cbica/projects/spatiotemp_dev_plasticity/software/rotate_parcellation/R/rotate.parcellation.R")
source("/cbica/projects/spatiotemp_dev_plasticity/software/rotate_parcellation/R/perm.sphere.p.R")
glasser.coords <- read.table("/cbica/projects/spatiotemp_dev_plasticity/software/rotate_parcellation/sphere_HCP.txt", header=F) #coordinates of glasser parcel centroids on the freesurfer sphere
perm.id.full <- rotate.parcellation(coord.l = as.matrix(glasser.coords[1:180,]), coord.r = as.matrix(glasser.coords[181:360,]), nrot = 10000) #rotate the glasser parcellation 10,000 times on the freesurfer sphere to generate spatial nulls for spin-based permutation significance testing 
saveRDS(perm.id.full, "/cbica/projects/spatiotemp_dev_plasticity/software/rotate_parcellation/glasser_sphericalrotations_N10000.rds")
perm.id.full <- readRDS("/cbica/projects/spatiotemp_dev_plasticity/software/rotate_parcellation/glasser_sphericalrotations_N10000.rds") #10,000 spatial null spins

GAM Results

gam.age.glasser <- read.csv("/cbica/projects/spatiotemp_dev_plasticity/FluctuationAmplitude/GAMRESULTS/fluctuationamplitude_age_statistics_glasser.csv") #GAM age smooth statistics, generated with fitGAMs_fluctuationamplitude_age.R

#create df.dev with GAM statistics, S-A axis, plasticity measures, and SNR mask
df.list <- list(gam.age.glasser, S.A.axis, myelin, SNR.mask) #dfs to merge
df.dev <- Reduce(function(x,y) merge(x,y, all=TRUE, sort=F), df.list) 
df.dev <- df.dev %>% select(label, orig_parcelname, everything()) #order columns
cols = c(3:15)    
df.dev[,cols] = apply(df.dev[,cols], 2, function(x) as.numeric(as.character(x))) #format numerics

#create df.dev.spin formatted for spatial permutation testing
df.dev.spin <- rbind(df.dev[181:360,], df.dev[1:180,]) #format df as left hemisphere -> right hemisphere for spin tests
df.dev.spin$GAM.age.partialR2[df.dev.spin$SNR.mask == 0] <- NA #format data for spin tests by assigning NA to low SNR parcels, treating them like the medial wall
df.dev.spin$minage.decrease[df.dev.spin$SNR.mask == 0] <- NA 

df.dev <- df.dev %>% filter(SNR.mask != 0) #include only high SNR parcels (N=336) in analyses
df.dev$SA.axis.bin <- as.numeric(cut2(df.dev$SA.axis, g=10)) #divide the S-A axis into 10 ranked bins
gam.fitted.glasser <- read.csv("/cbica/projects/spatiotemp_dev_plasticity/FluctuationAmplitude/GAMRESULTS/fluctuationamplitude_age_predictedfits_glasser.csv") #GAM predicted fits, generated with fitGAMs_fluctuationamplitude_age.R
gam.fitted.glasser <- inner_join(gam.fitted.glasser, df.dev, by="label", sort = F)
gam.smoothestimates.glasser <- read.csv("/cbica/projects/spatiotemp_dev_plasticity/FluctuationAmplitude/GAMRESULTS/fluctuationamplitude_age_smoothestimates_glasser.csv") #GAM smooth estimates, generated with fitGAMs_fluctuationamplitude_age.R
gam.smoothestimates.glasser <- inner_join(gam.smoothestimates.glasser, df.dev, by="label", sort = F)
gam.derivatives.glasser <- read.csv("/cbica/projects/spatiotemp_dev_plasticity/FluctuationAmplitude/GAMRESULTS/fluctuationamplitude_age_derivatives_glasser.csv") #GAM true model age smooth derivatives, generated with fitGAMs_fluctuationamplitude_age.R
gam.derivatives.glasser <- left_join(gam.derivatives.glasser, S.A.axis, by="label", sort = F)
gam.derivatives.glasser <- left_join(gam.derivatives.glasser, SNR.mask, by="label", sort = F)
gam.derivatives.glasser <- gam.derivatives.glasser %>% filter(SNR.mask != 0) 
gam.env.glasser <- read.csv("/cbica/projects/spatiotemp_dev_plasticity/FluctuationAmplitude/GAMRESULTS/fluctuationamplitude_environment_statistics_glasser.csv") #GAM environment statistics, generated with fitGAMs_fluctuationamplitude_environment.R

#create df.env with GAM results, S-A axis, and SNR mask
df.list <- list(gam.env.glasser, S.A.axis, SNR.mask) #dfs to merge
df.env <- Reduce(function(x,y) merge(x,y, all=TRUE, sort=F), df.list) 

#create df.env.spin formatted for spatial permutation testing
df.env.spin <- rbind(df.env[181:360,], df.env[1:180,]) #format df as left hemisphere -> right hemisphere for spin tests
df.env.spin$GAM.env.tvalue[df.env.spin$SNR.mask == 0] <- NA #format data for spin tests by assigning NA to low SNR parcels, treating them like the medial wall
df.env.spin$GAM.env.edu.tvalue[df.env.spin$SNR.mask == 0] <- NA

df.env <- df.env %>% filter(SNR.mask != 0) #include only high SNR parcels (N=336) in analyses

Age-Dependent Changes in Spontaneous Fluctuations Vary Across the Cortex

Cortical smooth functions

gam.smoothestimates.glasser.lh <- gam.smoothestimates.glasser[33401:67200,]
ggplot(gam.smoothestimates.glasser.lh,aes(age,est,group=index,color=GAM.age.partialR2)) + 
  geom_line(size=.8, alpha = .8) + 
  paletteer::scale_color_paletteer_c("pals::ocean.matter", direction = -1, limits = c(-0.07, .03), oob = squish) +
  theme_classic() +
  labs(x = "\nAge", y = "Fluctuation Amplitude (zero centered)\n" ) +
  theme(legend.position = "none") +
  theme(axis.text = element_text(size=15, family = "Arial", color = c("black")), axis.title = element_text(size=15, family = "Arial", color = c("black"))) +
  scale_x_continuous(breaks=c(8, 10, 12, 14, 16, 18, 20, 22), expand = c(0,.45))

Region-specific developmental trajectories and derivatives

Region V1

V1.smooth <- gam.fitted.glasser %>% filter(label == "rh_R_V1")
V1.derivatives <- gam.derivatives.glasser %>% filter(label == "rh_R_V1")

ggplot(data = fluctuations.glasser.pnc, aes(x = age, y = rh_R_V1)) +
  geom_point(color="#FBDC9D") +
  geom_line(data = V1.smooth, aes(x = age, y = fitted), color="#F8B57B", size=2) +
  geom_ribbon(data = V1.smooth, aes(x = age, y = fitted,  ymin = lower, ymax = upper) ,alpha = .7, linetype = 0, fill="#F8B87D",) +
  labs(x="\nAge", y="Fluctuation Amplitude\n") +
  theme_classic() +
  theme(
  axis.text = element_text(size=15, family = "Arial", color = c("black")),
  axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
  axis.title.y=element_text(size=15, family ="Arial", color = c("black"))) +
  scale_x_continuous(breaks=c(8, 10, 12, 14, 16, 18, 20, 22), expand = c(0,.45)) +
  scale_y_continuous(breaks=c(0.5, 1, 1.5, 2, 2.5, 3))

ggplot(data = V1.derivatives) + 
  geom_tile(aes(x = age, y = .1, fill = significant.derivative)) +
  scale_fill_gradient(low = "#EFAF77", high = "#FFE0A1", limits = c(-0.0323371,-.01), na.value = "white") +
  labs(x="\nAge", fill = "Derivative") + 
  theme_classic() +
  theme(axis.title.y = element_blank(),
          axis.text.y = element_blank(),
          axis.line = element_blank(),
          axis.ticks.y = element_blank(),
          axis.title.x=element_text(size=15, family ="Arial", color = c("black")))

Region p24pr

p24pr.smooth <- gam.fitted.glasser %>%filter(label == "rh_R_p24pr")
p24pr.derivatives <- gam.derivatives.glasser %>% filter(label == "rh_R_p24pr")

ggplot(data = fluctuations.glasser.pnc, aes(x = age, y = rh_R_p24pr)) +
  geom_point(color="#DF5E54") +
  geom_line(data = p24pr.smooth, aes(x = age, y = fitted), color="#D54D55", size=2) +
  geom_ribbon(data = p24pr.smooth, aes(x = age, y = fitted,  ymin = lower, ymax = upper) ,alpha = .7, linetype = 0, fill="#D54D55",) +
  labs(x="\nAge", y="Fluctuation Amplitude\n") +
  theme_classic() +
  theme(
  axis.text = element_text(size=15, family = "Arial", color = c("black")),
  axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
  axis.title.y=element_text(size=15, family ="Arial", color = c("black"))) +
  scale_x_continuous(breaks=c(8, 10, 12, 14, 16, 18, 20, 22), expand = c(0,.45)) +
  scale_y_continuous(breaks=c(0.5, 1, 1.5, 2, 2.5, 3))

ggplot(data = p24pr.derivatives) + 
  geom_tile(aes(x = age, y = .5, fill = significant.derivative)) +
  scale_fill_gradient(high = alpha("#DF5E54",0.7), low = "#D54D55", na.value = "white", limits = c(min(p24pr.derivatives$significant.derivative),-.001)) +
  labs(x="\nAge", fill = "Derivative") + 
  theme_classic() +
  theme(axis.title.y = element_blank(),
          axis.text.y = element_blank(),
          axis.line = element_blank(),
          axis.ticks.y = element_blank(),
          axis.title.x=element_text(size=15, family ="Arial", color = c("black")))

Region IFSa

#rh_R_9m
IFSa.smooth <- gam.fitted.glasser %>%filter(label == "lh_L_IFSa")
IFSa.derivatives <- gam.derivatives.glasser %>% filter(label == "lh_L_IFSa")

ggplot(data = fluctuations.glasser.pnc, aes(x = age, y = lh_L_IFSa)) +
  geom_point(color="#6E195E") +
  geom_line(data = IFSa.smooth, aes(x = age, y = fitted), color="#381043", size=2) +
  geom_ribbon(data = IFSa.smooth, aes(x = age, y = fitted,  ymin = lower, ymax = upper) ,alpha = .7, linetype = 0, fill="#381043",) +
  labs(x="\nAge", y="Fluctuation Amplitude\n") +
  theme_classic() +
  theme(
  axis.text = element_text(size=15, family = "Arial", color = c("black")),
  axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
  axis.title.y=element_text(size=15, family ="Arial", color = c("black"))) +
  scale_x_continuous(breaks=c(8, 10, 12, 14, 16, 18, 20, 22), expand = c(0,.45)) +
  scale_y_continuous(breaks=c(0.5, 1, 1.5, 2, 2.5, 3))

ggplot(data = IFSa.derivatives) + 
  geom_tile(aes(x = age, y = .5, fill = significant.derivative)) +
  scale_fill_gradient2(high = "#6E195E", low = "#cf93c4", midpoint = 0, mid = "white", na.value = "white") +
  labs(x="\nAge", fill = "Derivative") + 
  theme_classic() +
  theme(axis.title.y = element_blank(),
          axis.text.y = element_blank(),
          axis.line = element_blank(),
          axis.ticks.y = element_blank(),
          axis.title.x=element_text(size=15, family ="Arial", color = c("black")))

Spatiotemporal Development of Fluctuation Amplitude Follows Intracortical Myelin Growth

Spatial development

Fluctuation amplitude age effects map

ggseg(.data = df.dev, atlas = "glasser", mapping=aes(fill=GAM.age.partialR2), position = c("stacked"), hemisphere = c("right")) + 
  theme_void() + 
  labs(fill="Age Effect") +
  paletteer::scale_fill_paletteer_c("pals::ocean.matter", direction = -1, limits = c(-0.07, .03), oob = squish)

T1/T2 ratio age effects map

ggseg(.data = df.dev, atlas = "glasser", mapping=aes(fill=myelin.partialR2), position = c("stacked"), hemisphere = "right") + 
  theme_void() + 
  labs(fill="Age Effect") +
  paletteer::scale_fill_paletteer_c("pals::ocean.matter", direction = 1, limits = c(0.0,0.3), oob = squish) 
## merging atlas and data by 'label'
## Warning: Some data not merged properly. Check for naming errors in data:
##    atlas type  hemi  side  region label    orig_parcelname GAM.age.Fvalue
##    <chr> <chr> <chr> <chr> <chr>  <chr>    <chr>                    <dbl>
##  1 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V1  L_V1_ROI                 27.6 
##  2 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_MST L_MST_ROI                14.7 
##  3 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V6  L_V6_ROI                  7.44
##  4 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V2  L_V2_ROI                 33.4 
##  5 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V3  L_V3_ROI                 27.8 
##  6 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V4  L_V4_ROI                 45.6 
##  7 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V8  L_V8_ROI                 29.3 
##  8 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_4   L_4_ROI                  20.6 
##  9 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_3b  L_3b_ROI                 16.2 
## 10 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_FEF L_FEF_ROI                13.3 
## # … with 159 more rows, and 13 more variables: GAM.age.pvalue <dbl>,
## #   GAM.age.partialR2 <dbl>, Anova.age.pvalue <dbl>, age.onsetchange <dbl>,
## #   age.peakchange <dbl>, minage.decrease <dbl>, maxage.increase <dbl>,
## #   age.maturation <dbl>, SA.axis <dbl>, myelin.partialR2 <dbl>,
## #   myelin.maxdev <dbl>, SNR.mask <dbl>, SA.axis.bin <dbl>

Correlation (rho + p.spin) between fluctuation amplitude age effects and T1/T2 ratio age effects

cor.test(df.dev$GAM.age.partialR2, df.dev$myelin.partialR2, method=c("spearman"))
## 
##  Spearman's rank correlation rho
## 
## data:  df.dev$GAM.age.partialR2 and df.dev$myelin.partialR2
## S = 10557746, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##        rho 
## -0.6699693
perm.sphere.p(df.dev.spin$GAM.age.partialR2, df.dev.spin$myelin.partialR2, perm.id.full,corr.type='spearman')
## [1] 0.00045

Correlation plot

ggplot(df.dev, aes(x=myelin.partialR2, y=GAM.age.partialR2, fill = myelin.partialR2)) + 
geom_point(color = "white",shape=21, size=3.7) +
paletteer::scale_fill_paletteer_c("pals::ocean.matter", direction = 1, limits = c(0.0,0.3), oob = squish) +
labs(x="\nMyelin Age Effect", y="Fluctuation Amplitude Age Effect\n") +
geom_smooth(method='lm', se=TRUE, fill=alpha(c("gray70"),.7), col="black") +
theme(
axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
axis.title.y=element_text(size=15, family ="Arial", color = c("black")),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, family ="Arial", color = c("black")),
panel.background=element_blank(),
legend.position = "none")
## `geom_smooth()` using formula 'y ~ x'

Temporal development

Fluctuation amplitude age of decrease onset

ggseg(.data = df.dev, atlas = "glasser", mapping=aes(fill=minage.decrease), position = c("stacked"), hemisphere = "right") + 
  theme_void() + 
  labs(fill="Age Decrease Onset") +
  paletteer::scale_fill_paletteer_c("pals::ocean.matter", direction = -1, limits = c(9, 17.5), oob = squish)

T1/T2 ratio age of maximal increase

ggseg(.data = df.dev, atlas = "glasser", mapping=aes(fill=myelin.maxdev), position = c("stacked"), hemisphere = "right") + 
  theme_void() + 
  labs(fill="Age Maximal Increase") +
  paletteer::scale_fill_paletteer_c("pals::ocean.matter", direction = -1, limits = c(10,17.5), oob = squish) 
## merging atlas and data by 'label'
## Warning: Some data not merged properly. Check for naming errors in data:
##    atlas type  hemi  side  region label    orig_parcelname GAM.age.Fvalue
##    <chr> <chr> <chr> <chr> <chr>  <chr>    <chr>                    <dbl>
##  1 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V1  L_V1_ROI                 27.6 
##  2 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_MST L_MST_ROI                14.7 
##  3 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V6  L_V6_ROI                  7.44
##  4 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V2  L_V2_ROI                 33.4 
##  5 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V3  L_V3_ROI                 27.8 
##  6 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V4  L_V4_ROI                 45.6 
##  7 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_V8  L_V8_ROI                 29.3 
##  8 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_4   L_4_ROI                  20.6 
##  9 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_3b  L_3b_ROI                 16.2 
## 10 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_FEF L_FEF_ROI                13.3 
## # … with 159 more rows, and 13 more variables: GAM.age.pvalue <dbl>,
## #   GAM.age.partialR2 <dbl>, Anova.age.pvalue <dbl>, age.onsetchange <dbl>,
## #   age.peakchange <dbl>, minage.decrease <dbl>, maxage.increase <dbl>,
## #   age.maturation <dbl>, SA.axis <dbl>, myelin.partialR2 <dbl>,
## #   myelin.maxdev <dbl>, SNR.mask <dbl>, SA.axis.bin <dbl>

Correlation (rho + p.spin) between fluctuation amplitude age of decrease onset and T1/T2 ratio max age of increase

#All regions
cor.test(df.dev$minage.decrease, df.dev$myelin.maxdev, method=c("spearman"))
## Warning in cor.test.default(df.dev$minage.decrease, df.dev$myelin.maxdev, :
## Cannot compute exact p-value with ties
## 
##  Spearman's rank correlation rho
## 
## data:  df.dev$minage.decrease and df.dev$myelin.maxdev
## S = 1677077, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##       rho 
## 0.6382729
perm.sphere.p(df.dev.spin$minage.decrease, df.dev.spin$myelin.maxdev, perm.id.full,corr.type='spearman')
## [1] 0.00125
#Age of decrease onset > 8.5
late.decrease <- df.dev %>% filter(minage.decrease > 8.5)
cor.test(late.decrease$minage.decrease, late.decrease$myelin.maxdev, method=c("spearman"))
## Warning in cor.test.default(late.decrease$minage.decrease,
## late.decrease$myelin.maxdev, : Cannot compute exact p-value with ties
## 
##  Spearman's rank correlation rho
## 
## data:  late.decrease$minage.decrease and late.decrease$myelin.maxdev
## S = 349670, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##       rho 
## 0.6402458
late.decrease.spin <- df.dev.spin
late.decrease.spin$minage.decrease[late.decrease.spin$minage.decrease < 8.5] <- NA
perm.sphere.p(late.decrease.spin$minage.decrease, late.decrease.spin$myelin.maxdev, perm.id.full,corr.type='spearman')
## [1] 0.01565

Correlation plot

ggplot(df.dev, aes(x=myelin.maxdev, y=minage.decrease, fill = myelin.maxdev)) + 
geom_point(color = "white",shape=21, size=3.7) +
paletteer::scale_fill_paletteer_c("pals::ocean.matter",  direction = -1, limits = c(10,17.5), oob = squish) +
labs(x="\nMyelin Age of Maximal Increase", y="Fluctuation Amplitude Age of Decrease Onset\n") +
geom_smooth(method='lm', se=TRUE, fill=alpha(c("gray70"),.7), col="black") +
theme(
axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
axis.title.y=element_text(size=15, family ="Arial", color = c("black")),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, family ="Arial", color = c("black")),
panel.background=element_blank(),
legend.position = "none")

ggplot(late.decrease, aes(x=myelin.maxdev, y=minage.decrease, fill = myelin.maxdev)) + 
geom_point(color = "white",shape=21, size=3.7) +
paletteer::scale_fill_paletteer_c("pals::ocean.matter",  direction = -1, limits = c(10,17.5), oob = squish) +
labs(x="\nMyelin Age of Maximal Increase", y="Fluctuation Amplitude Age of Decrease Onset\n") +
geom_smooth(method='lm', se=TRUE, fill=alpha(c("gray70"),.7), col="black") +
theme(
axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
axis.title.y=element_text(size=15, family ="Arial", color = c("black")),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, family ="Arial", color = c("black")),
panel.background=element_blank(),
legend.position = "none")

Fluctuation Amplitude Development Unfolds along the Hierarchical Sensorimotor-Association Axis

Sensorimotor-association axis

ggseg(.data = df.dev, atlas = "glasser", mapping=aes(fill=SA.axis), position = c("stacked")) + 
  theme_void() + 
  theme(legend.position = "none") +
  scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 180)

Age effects along the S-A Axis

Correlation (rho + p.spin) between fluctuation amplitude age effects and the sensorimotor-association axis

cor.test(df.dev$GAM.age.partialR2, df.dev$SA.axis, method=c("spearman"))
## 
##  Spearman's rank correlation rho
## 
## data:  df.dev$GAM.age.partialR2 and df.dev$SA.axis
## S = 2934378, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##       rho 
## 0.5358554
perm.sphere.p(df.dev.spin$GAM.age.partialR2, df.dev.spin$SA.axis, perm.id.full,corr.type='spearman')
## [1] 0.00215

Correlation plot

ggplot(df.dev, aes(x=SA.axis, y=GAM.age.partialR2, fill = SA.axis)) + 
geom_point(color = "white",shape=21, size=4.5) +
scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 180) +
labs(x="\nSensorimtor-Association Axis Rank", y="Fluctuation Amplitude Age Effect\n") +
geom_smooth(method='lm', se=TRUE, fill=alpha(c("gray70"),.7), col="black") +
theme(
axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
axis.title.y=element_text(size=15, family ="Arial", color = c("black")),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, family ="Arial", color = c("black")),
panel.background=element_blank(),
legend.position = "none")

Temporal change along the S-A Axis

Correlation (rho + p.spin) between fluctuation amplitude age of decrease onset and the sensorimotor-association axis

cor.test(df.dev$minage.decrease, df.dev$SA.axis, method=c("spearman"))
## Warning in cor.test.default(df.dev$minage.decrease, df.dev$SA.axis, method =
## c("spearman")): Cannot compute exact p-value with ties
## 
##  Spearman's rank correlation rho
## 
## data:  df.dev$minage.decrease and df.dev$SA.axis
## S = 2477911, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##       rho 
## 0.4655417
perm.sphere.p(df.dev.spin$minage.decrease, df.dev.spin$SA.axis, perm.id.full,corr.type='spearman')
## [1] 0.0388
cor.test(late.decrease$minage.decrease, late.decrease$SA.axis, method=c("spearman"))
## Warning in cor.test.default(late.decrease$minage.decrease,
## late.decrease$SA.axis, : Cannot compute exact p-value with ties
## 
##  Spearman's rank correlation rho
## 
## data:  late.decrease$minage.decrease and late.decrease$SA.axis
## S = 314027, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##       rho 
## 0.6769174
perm.sphere.p(late.decrease.spin$minage.decrease, late.decrease.spin$SA.axis, perm.id.full,corr.type='spearman')
## [1] 0.0011

Correlation plot

ggplot(df.dev, aes(x=SA.axis, y=minage.decrease, fill = SA.axis)) + 
geom_point(color = "white",shape=21, size=3.7) +
scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 180) +
labs(x="\nSensorimtor-Association Axis", y="Fluctuation Amplitude Age of Decrease Onset\n") +
geom_smooth(method='lm', se=TRUE, fill=alpha(c("gray70"),.7), col="black") +
theme(
axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
axis.title.y=element_text(size=15, family ="Arial", color = c("black")),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, family ="Arial", color = c("black")),
panel.background=element_blank(),
legend.position = "none")

ggplot(late.decrease, aes(x=SA.axis, y=minage.decrease, fill = SA.axis)) + 
geom_point(color = "white",shape=21, size=3.7) +
scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 180) +
labs(x="\nSensorimtor-Association Axis", y="Fluctuation Amplitude Age of Decrease Onset\n") +
geom_smooth(method='lm', se=TRUE, fill=alpha(c("gray70"),.7), col="black") +
theme(
axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
axis.title.y=element_text(size=15, family ="Arial", color = c("black")),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, family ="Arial", color = c("black")),
panel.background=element_blank(),
legend.position = "none")

Sensorimotor-association axis binned GAM smooths

gam.smoothestimates.glasser$age <- round(gam.smoothestimates.glasser$age, 3)

meansmooth.bins <- gam.smoothestimates.glasser %>% group_by(age, SA.axis.bin) %>% do(est.mean = mean(.$est)) %>% unnest(cols = c(est.mean)) #calculate the average smooth estimate at each age within each of the 10 S-A axis bins
ggplot(data = meansmooth.bins, aes(x = age, y = est.mean, group = SA.axis.bin, color = as.factor(SA.axis.bin))) +
  geom_line(size = 2.5) +
  scale_color_manual(values = c("#FFC228", "#FFCA4E", "#FFD16A", "#FFDA89", "#FFE6B0", "#D7BCDA", "#BE93C4", "#9859A4", "#863E95", "#6F1382")) +
  theme_classic() +
  labs(x = "\nAge", y = "Fluctuation Amplitude (zero centered)\n", color = "Sensorimotor-Association Axis Bin") +
  theme(axis.text = element_text(size=15, family = "Arial", color = c("black")), axis.title = element_text(size=15, family = "Arial", color = c("black"))) +
  theme(legend.position = c(.585, 0.94), legend.direction = "horizontal", legend.text = element_text(size=13), legend.title = element_text(size=12, family = "Arial", color = c("black"))) +
  scale_x_continuous(breaks=c(8, 10, 12, 14, 16, 18, 20, 22), expand = c(0,.45))

Principal developmental gradient

Principal component analysis to identify the spatial axis that captures the most variance in regional fluctuation amplitude developmental trajectories

gam.smoothestimates.glasser <- gam.smoothestimates.glasser %>% select(label, age, est)
gam.smoothestimates.long <- gam.smoothestimates.glasser %>% pivot_wider(names_from="label",values_from="est")
gam.smoothestimates.long <- gam.smoothestimates.long %>% select(-age) #ages by regions matrix for PCA

smooths.pca <- prcomp(gam.smoothestimates.long) #PCA

Variance explained by PC1

summary(smooths.pca)$importance[2,1]
## [1] 0.87026

Developmental PC1 correlation with the sensorimotor-association axis

PC1 <- as.data.frame(smooths.pca$rotation[,1]) #first principal component loadings
PC1$label <- row.names(PC1) 
colnames(PC1) <- c("PC1","label")
PC1$ranked <- rank(PC1$PC1)

cor.test(PC1$PC1, df.dev$SA.axis, method=c("spearman"))
## 
##  Spearman's rank correlation rho
## 
## data:  PC1$PC1 and df.dev$SA.axis
## S = 1896048, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##      rho 
## 0.700093
df.dev.spin <- left_join(df.dev.spin, PC1, by="label")
perm.sphere.p(df.dev.spin$PC1, df.dev.spin$SA.axis, perm.id.full,corr.type='spearman')
## [1] 0

Correlation plot

PC1 <- left_join(PC1, S.A.axis, by="label")
ggplot(PC1, aes(x=SA.axis, y=PC1, fill = SA.axis)) + 
geom_point(color = "white",shape=21, size=3.7) +
scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 180) +
labs(x="\nSensorimtor-Association Axis", y="Development PC1\n") +
geom_smooth(method='lm', se=TRUE, fill=alpha(c("gray70"),.7), col="black") +
theme(
axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
axis.title.y=element_text(size=15, family ="Arial", color = c("black")),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, family ="Arial", color = c("black")),
panel.background=element_blank(),
legend.position = "none")

PCA brain map

ggseg(.data = PC1, atlas = "glasser", mapping=aes(fill=PC1), position = c("stacked")) + theme_void() + 
scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = -0.055, limits = c(-0.08,0.008), oob = squish) + theme(legend.position = "none")
## merging atlas and data by 'label'
## Warning: Some data not merged properly. Check for naming errors in data:
##   atlas type  hemi  side  region label        PC1 ranked orig_parcelname SA.axis
##   <chr> <chr> <chr> <chr> <chr>  <chr>      <dbl>  <dbl> <chr>             <dbl>
## 1 <NA>  <NA>  <NA>  <NA>  <NA>   lh_L_10pp 0.0175    319 L_10pp_ROI          282

The Neurodevelopmental Hierarchy is Dominant through Adolescence

Age-specific fluctuation amplitude derivatives

Regional derivative by age plot

gam.derivatives.glasser.lh <- gam.derivatives.glasser[33401:67200,]
ggplot(gam.derivatives.glasser.lh,aes(age,derivative,group=index,color=SA.axis)) + 
  geom_line(size=1, alpha = .7) + 
  scale_color_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "color", name = NULL, midpoint = 180) +
  theme_classic() +
  theme(legend.position = "none") +
  labs(x = "\nAge", y = "Derivative\n", color = "Sensorimotor-Association Axis Bin") +
  theme(axis.text = element_text(size=15, family = "Arial", color = c("black")), axis.title = element_text(size=15, family = "Arial", color = c("black"))) +
  scale_x_continuous(breaks=c(8, 10, 12, 14, 16, 18, 20, 22), expand = c(0,.45))

gam.derivatives.glasser$age <- round(gam.derivatives.glasser$age, 5)

Age 10 derivative map

age10.derivs <- gam.derivatives.glasser %>% filter(age == 10.03015)
age10.derivs$derivative <- rank(age10.derivs$derivative)
ggseg(.data = age10.derivs, atlas = "glasser", mapping=aes(fill=derivative), position = c("stacked"), hemisphere = "right") + 
  theme_void() + 
  scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 168)

rm(age10.derivs)

Age 15 derivative map

age15.derivs <- gam.derivatives.glasser %>% filter(age == 15.02429)
age15.derivs$derivative <- rank(age15.derivs$derivative)
ggseg(.data = age15.derivs, atlas = "glasser", mapping=aes(fill=derivative), position = c("stacked"), hemisphere = "right") + 
  theme_void() + 
  scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 168)

rm(age15.derivs)

Age 20 derivative map

age20.derivs <- gam.derivatives.glasser %>% filter(age == 20.01843)
age20.derivs$derivative <- rank(age20.derivs$derivative)
ggseg(.data = age20.derivs, atlas = "glasser", mapping=aes(fill=derivative), position = c("stacked"), hemisphere = "right") + 
  theme_void() + 
  scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 168)

rm(age20.derivs)

Derivative by sensorimotor-association axis plot

ggplot(data=gam.derivatives.glasser,aes(x = age,y = SA.axis, group = index)) +
  geom_line(aes(color=derivative), size=1) +
  #paletteer::scale_color_paletteer_c("grDevices::RdYlBu", direction = -1, limits=c(-0.03,0.03), oob = squish) +
  paletteer::scale_color_paletteer_c("grDevices::Sunset", direction = -1, limits=c(-0.03,0.03), oob = squish) +
  #paletteer::scale_color_paletteer_c("grDevices::Inferno", direction = -1, limits=c(-0.03,0.03), oob = squish) +
  theme_classic() +
  ylab("Sensorimotor-Association Axis Rank\n") +
  xlab("\nAge") +
  labs(color="Amplitude change per year\n") +
  theme(axis.text.x = element_text(size = 15, color = c("black"))) +
  theme(axis.text.y = element_text(size = 15, color = c("black"))) +
  scale_x_continuous(breaks=c(8, 10, 12, 14, 16, 18, 20, 22), expand = c(0,.45))

Age-specific correlation between derivates and the S-A Axis

Posterior derivative - sensorimotor-association axis correlations, N=10,000

deriv.SAaxis.posteriorcorrs <- read.table("/cbica/projects/spatiotemp_dev_plasticity/FluctuationAmplitude/GAMRESULTS/SAaxis_posteriorderivative_correlation_byage_glasser.csv", header = F, sep = ",") #get the regional derivative-regional axis rank correlation at each age for all draws from the posterior distribution
colnames(deriv.SAaxis.posteriorcorrs) <- sprintf("draw%s",seq(from = 1, to = ncol(deriv.SAaxis.posteriorcorrs)))
deriv.SAaxis.posteriorcorrs <- cbind(deriv.SAaxis.posteriorcorrs, (gam.derivatives.glasser %>% group_by(age) %>% group_keys()))

Age of maximal sensorimotor-association axis correlation: posterior median value + 95% credible interval

age.max.corr <- deriv.SAaxis.posteriorcorrs %>% #find the age at which the correlation is largest for each draw
    summarise(across(contains("draw"),
                     .fns = function(x){
                       round(deriv.SAaxis.posteriorcorrs$age[which.max(x)],4)
                     }))
age.max.corr <- t(age.max.corr)
age.max.corr.median <- median(age.max.corr) #median age #bayes
age.max.corr.CI <- quantile(age.max.corr, probs = c(0.025, 0.975)) #compute the credible interval based on all draws
age.max.corr.lower <- age.max.corr.CI[[1]]
age.max.corr.upper <- age.max.corr.CI[[2]]
age.max.corr.median
## [1] 15.0243
age.max.corr.CI
##    2.5%   97.5% 
## 14.6516 15.3224

Sensorimotor-association axis correlation value at each age: posterior median values + 95% credible interval

deriv.SAaxis.posteriorcorrs <- deriv.SAaxis.posteriorcorrs %>% select(contains("draw"))

deriv.SAaxis.mediancorr <- apply(deriv.SAaxis.posteriorcorrs, 1, function(x){median(x)}) #median correlation value at each age

cor.credible.intervals <- apply(deriv.SAaxis.posteriorcorrs, 1, function(x){quantile(x, probs = c(0.025, 0.975))}) #compute the credible interval for the correlation value at each age based on all draws
cor.credible.intervals <- t(cor.credible.intervals)
cor.credible.intervals <- as.data.frame(cor.credible.intervals)
cor.credible.intervals <- cbind(cor.credible.intervals, (gam.derivatives.glasser %>% group_by(age) %>% group_keys()))
cor.credible.intervals$SA.correlation <- deriv.SAaxis.mediancorr 
colnames(cor.credible.intervals) <- c("lower","upper","age","SA.correlation")

Window of null sensorimotor-association axis correlation

cor.credible.intervals$max.corr.CI <- (cor.credible.intervals$age > age.max.corr.lower & cor.credible.intervals$age < age.max.corr.upper) #add a column that indicates whether each age is included in the age of maximal correlation credible interval (T/F)
cor.credible.intervals$max.cor.window <- cor.credible.intervals$age*cor.credible.intervals$max.corr.CI #add a column that only includes ages in this interval
cor.credible.intervals$max.cor.window[cor.credible.intervals$max.cor.window == 0] <- NA
                        
cor.credible.intervals$zero.corr.CI <- (cor.credible.intervals$lower < 0 & cor.credible.intervals$upper > 0) #add a column that indicates whether the credible interval for the correlation includes 0
cor.credible.intervals$zero.corr.window <- cor.credible.intervals$age*cor.credible.intervals$zero.corr.CI #add a column that only includes ages in the zero window
cor.credible.intervals$zero.corr.window[cor.credible.intervals$zero.corr.window == 0] <- NA
min(cor.credible.intervals$age[cor.credible.intervals$zero.corr.CI == TRUE])
## [1] 19.12395
max(cor.credible.intervals$age[cor.credible.intervals$zero.corr.CI == TRUE])
## [1] 22.47822

Sensorimotor-association axis development sliding window plot

ggplot(cor.credible.intervals, aes(x = age, y = SA.correlation, ymin = lower, ymax = upper)) + 
geom_line(size=1.5) +
geom_ribbon(alpha=.2, fill=c("grey30")) +
labs(x="\nAge", y="Developmental Alignment the Axis\n") +
geom_ribbon(aes(x = max.cor.window, y = SA.correlation), fill="#F4A674") +
geom_ribbon(aes(x = zero.corr.window, y = SA.correlation), fill="#952162") +  
theme(
axis.title.x=element_text(size=15, color = "black"),
axis.title.y=element_text(size=15, color = "black"),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, color = "black"),
panel.background=element_blank(),legend.position = "none") +
scale_y_continuous(breaks=c(0,0.2,0.4,0.6)) +
scale_x_continuous(breaks=c(8, 10, 12, 14, 16, 18, 20, 22), limits = c(8, 23), expand = c(0,0))

Fluctuation Amplitude Variability is Linked to Variability in the Developmental Environment

Significant associations between fluctuation amplitude and environmental factors

Number of cortical regions showing significant environment effects

  • original model
sum(p.adjust(df.env$Anova.env.pvalue, method=c("fdr")) < 0.05) 
## [1] 141
  • controlling for parental education
sum(p.adjust(df.env$Anova.env.edu.pvalue, method=c("fdr")) < 0.05)
## [1] 106

Percent of cortical regions showing significant environment effects

  • original model
(sum(p.adjust(df.env$Anova.env.pvalue, method=c("fdr")) < 0.05))/(nrow(df.env))*100 
## [1] 41.96429
  • controlling for parental education
(sum(p.adjust(df.env$Anova.env.edu.pvalue, method=c("fdr")) < 0.05))/(nrow(df.env))*100 
## [1] 31.54762

Across-cortex environment effect size and direction

ggseg(.data = df.env, atlas = "glasser", mapping=aes(fill=GAM.env.tvalue), position = c("stacked")) + 
  theme_void() + 
  labs(fill="Age Effect") +
  scale_fill_gradient2(low= "#F4A674", mid = "white", high = "#952162", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 0, limits=c(-4,4), oob=squish)

Correlation between environmental effects and the sensorimotor-association axis

  • original model
#all regions
cor.test(df.env$GAM.env.tvalue, df.env$SA.axis, method=c("spearman"))
## 
##  Spearman's rank correlation rho
## 
## data:  df.env$GAM.env.tvalue and df.env$SA.axis
## S = 3274572, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##       rho 
## 0.4820453
perm.sphere.p(df.env.spin$GAM.env.tvalue, df.env.spin$SA.axis, perm.id.full, corr.type='spearman')
## [1] 0
#significant regions
df.env.spin$GAM.env.tvalue.significant <- df.env.spin$GAM.env.tvalue*(p.adjust(df.env.spin$Anova.env.pvalue, method=c("fdr")) < 0.05)
df.env.spin$GAM.env.tvalue.significant[df.env.spin$GAM.env.tvalue.significant == 0] <- NA

cor.test(df.env.spin$GAM.env.tvalue.significant, df.env.spin$SA.axis, method=c("spearman"))
## 
##  Spearman's rank correlation rho
## 
## data:  df.env.spin$GAM.env.tvalue.significant and df.env.spin$SA.axis
## S = 204474, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##       rho 
## 0.5623229
perm.sphere.p(df.env.spin$GAM.env.tvalue.significant, df.env.spin$SA.axis, perm.id.full, corr.type='spearman')
## [1] 0

Correlation plot

ggplot(df.env, aes(x=SA.axis, y=GAM.env.tvalue, fill = SA.axis)) + 
geom_point(color = "white",shape=21, size=3.7) +
scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 180) +
labs(x="\nSensorimtor-Association Axis", y="Environment Effect\n") +
geom_smooth(method='lm', se=TRUE, fill=alpha(c("gray70"),.7), col="black") +
theme(
axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
axis.title.y=element_text(size=15, family ="Arial", color = c("black")),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, family ="Arial", color = c("black")),
panel.background=element_blank(),
legend.position = "none")

  • controlling for parental education
#all regions
cor.test(df.env$GAM.env.edu.tvalue, df.env$SA.axis, method=c("spearman"))
## 
##  Spearman's rank correlation rho
## 
## data:  df.env$GAM.env.edu.tvalue and df.env$SA.axis
## S = 2914840, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##       rho 
## 0.5389458
perm.sphere.p(df.env.spin$GAM.env.edu.tvalue, df.env.spin$SA.axis, perm.id.full, corr.type='spearman')
## [1] 0
#significant regions
df.env.spin$GAM.env.edu.tvalue.significant <- df.env.spin$GAM.env.edu.tvalue*(p.adjust(df.env.spin$Anova.env.edu.pvalue, method=c("fdr")) < 0.05)
df.env.spin$GAM.env.edu.tvalue.significant[df.env.spin$GAM.env.edu.tvalue.significant == 0] <- NA

cor.test(df.env.spin$GAM.env.edu.tvalue.significant, df.env.spin$SA.axis, method=c("spearman"))
## 
##  Spearman's rank correlation rho
## 
## data:  df.env.spin$GAM.env.edu.tvalue.significant and df.env.spin$SA.axis
## S = 83236, p-value < 2.2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##       rho 
## 0.5806434
perm.sphere.p(df.env.spin$GAM.env.edu.tvalue.significant, df.env.spin$SA.axis, perm.id.full, corr.type='spearman')
## [1] 0

Correlation plot

ggplot(df.env, aes(x=SA.axis, y=GAM.env.edu.tvalue, fill = SA.axis)) + 
geom_point(color = "white",shape=21, size=3.7) +
scale_fill_gradient2(low= "goldenrod1", mid = "white", high = "#6f1282", guide = "colourbar", aesthetics = "fill", name = NULL, midpoint = 180) +
labs(x="\nSensorimtor-Association Axis", y="Environment Effect\n") +
geom_smooth(method='lm', se=TRUE, fill=alpha(c("gray70"),.7), col="black") +
theme(
axis.title.x=element_text(size=15, family ="Arial", color = c("black")),
axis.title.y=element_text(size=15, family ="Arial", color = c("black")),
axis.line = element_line(color = "black"),
axis.text=element_text(size=15, family ="Arial", color = c("black")),
panel.background=element_blank(),
legend.position = "none")